cde wrote on 2022-07-24, 09:36:

Very good news! I did not have issues with Flatout2 with the defaults, but I'll turn it off nonetheless.

One thing I can confirm though is that my NV25 Ti4600 was not able to pass this test, while my NV28 card did fine... So perhaps the AGP on the KT7A is a bit anemic when it comes to raw power delivery (NV28 is probably a lot more power efficient than NV25)...

So maybe it's wiser to use something with an external connector (like an FX5800 or better, for example)... But I need to run more tests on other boards to confirm that this is unique to the KT7A.

I am also using an NV25, it's the Ti 4200 (clocked lower). This could explain the results indeed. Perhaps the Zalman cooler helps as well.

mockingbird wrote on 2022-07-15, 01:39:

I'm pretty sure I've traced down the Flatout2 instability to a setting in the KT7A's BIOS called "Enhance Chip Performance" under the "Softmenu III" category... I had to clear the CMOS on this board recently because I was experimenting with different heatsinks and CPUs, and the problem came back. So I started experimenting by enabling/disabling the performance enhancements in the BIOS. With this setting disabled, I completed the Desert Oil Field course without issue.

Well, I have to try this! Been using Radeon X800Pro with KT7A without any problems but any Geforce and Flatout2 throws BSOD or something else. I'll put my Geforce 6800GT in there to try that BIOS setting out.

Edit: Didnt help. 🙁

mockingbird wrote on 2022-07-15, 01:39:

I'm pretty sure I've traced down the Flatout2 instability to a setting in the KT7A's BIOS called "Enhance Chip Performance" under the "Softmenu III" category... I had to clear the CMOS on this board recently because I was experimenting with different heatsinks and CPUs, and the problem came back. So I started experimenting by enabling/disabling the performance enhancements in the BIOS. With this setting disabled, I completed the Desert Oil Field course without issue.

Having cleared the BIOS using the optimized values, the "Enhance Chip Performance" is by default disabled, I'm using the latest officiel BIOS for this motherboard. So it's possible you're seeing this issue because your custom BIOS has this option turned on by default?

cde wrote on 2022-08-04, 08:17:

mockingbird wrote on 2022-07-15, 01:39:

I'm pretty sure I've traced down the Flatout2 instability to a setting in the KT7A's BIOS called "Enhance Chip Performance" under the "Softmenu III" category... I had to clear the CMOS on this board recently because I was experimenting with different heatsinks and CPUs, and the problem came back. So I started experimenting by enabling/disabling the performance enhancements in the BIOS. With this setting disabled, I completed the Desert Oil Field course without issue.

Having cleared the BIOS using the optimized values, the "Enhance Chip Performance" is by default disabled, I'm using the latest officiel BIOS for this motherboard. So it's possible you're seeing this issue because your custom BIOS has this option turned on by default?

Yes, I acknowledge that Enhance Chip Performance is disabled in BIOS 3CX5 by default, so I'm not sure why it was enabled... Nevertheless, it did work fine with ti disabled... The KT7A did not like my NV25 GeForce (Ti4600), so I still have to test it on my other KT133A boards to see if it is the card, or the board. My hunch is it's the KT7A not liking the increased power draw (I think the card is OK)...

cde wrote on 2021-03-04, 07:17:

Thanks shamino, it is indeed a great option. Now I'd like to report on my (mis)adventures with two similar boards, the QDI KinetiZ 7E (KT133A based) and QDI KinetiZ 7B (KT133 based).

Another issue is the general lack of clocking/revolting options in the BIOS. The multiplier must be set manually with jumpers in accordance to the manual, with 12 being the highest. A CPU with unlocked multiplier is required, such as an Athlon XP mobile or Athlon XP before week 38 of 2003 (I tried an Athlon XP manufactured after this data and the system wouldn't boot). The maximum CPU frequency on the QDI KinetiZ 7E is 1600 MHz (133*12) and on the KinetiZ 7B 1200 MHz (100*12). The 7B in theory supports a 133 MHz FSB with a jumper but the system wouldn't boot even though the same CPU worked fine with FSB at 133 MHz on the 7E. (I don't think the KT133 supports a 133 MHz FSB anyway?)

Overall I wouldn't recommend the QDI KinetiZ 7B, the capacitor clearance and 100 MHz FSB makes it an inferior choice. The 7E is a bit better but not much; if you can live with the issues above then it may be considered.

Hi there

Sorry to bother you, but can i ask what have you tried on the Kinetiz 7B mb, when you tried to run it at 133MHz FSB? Like, what specific Athlon XP processor model did you use, did you change anything with the jumper settings when testing, etc.

I have a similiar motherboard, the QDI KinetiZ 7T and i too tried to put an Athlon XP with 133MHz FSB and it didn't post at all. I've been trying to find some information online, whether or not is possible to run 133 MHz Fsb on this motherboard and had no luck.

Unfortunately I can't remember exactly my tests, but from what I read online, that motherboard is not capable of running FSB at 133 MHz.

In general across all motherboards KT133A was always more reliable for hitting 133FSB, maybe even 140, whereas the KT133 stranded a lot of people at 115 or 124, with only a few lucky boards getting 133

BitWrangler wrote on 2022-09-29, 14:50:

In general across all motherboards KT133A was always more reliable for hitting 133FSB, maybe even 140, whereas the KT133 stranded a lot of people at 115 or 124, with only a few lucky boards getting 133

I wonder if there's really any benefit to the KT133A though, considering the tiny performance gain of a 133Mhz FSB on this legacy platform.

If a non 'a' KT133 board supports Barton CPUs, then I don't see any real difference between the two, considering all multiplier mods can be done on the pin side of the socket... One could theoretically set a 2800+ Barton to run at 21 * 100, instead of the default 166*12.5. We're already doing this anyways with the 2800+, because most versions are the 166Mhz version which are usually run at 16*133 (the 133Mhz FSB Bartons are not as easy to come by).

What remains to be seen though is if the original KT133 can handle SETMUL requests properly... It has been stated that on-the-fly multiplier changing was fixed in a later revision of the chip, but I'm not certain.

I've only used mobile multi setting on KT133A, but throwing capabilities of the chipset into confusion is that until late socket A, DDR-400 boards, manufacturers didn't take any notice of users wanting to use mobile features, so maybe casually blocked, or didn't specifically implement it's function. So you get a some do some don't situation. Ergo, a wide range of KT133 non-A boards would have to be tested before it was concluded whether the chipset allows it or not. Best guess would be non-A versions of boards where it has been seen to work on the KT133A, like the original A7V.

edit: another point of confusion is that some users reported motherboards "don't support" mobile, when they tested them with mobile unlocked desktop chips... what is happening there is that the desktop chips don't have a mobile multiplier set on the bridges, so their max mobile multi is 24X which will likely not boot unless you have a Tbred-B or Barton mobile on bus set to 100Mhz. However, other boards with mobile enabled boot at the set desktop multi, so "The mod worked, the CPU is fine". With a factory mobile chip, it has it's rated mobile multi set to something sensible and this doesn't happen on those boards.

cde wrote on 2022-09-29, 11:49:

Unfortunately I can't remember exactly my tests, but from what I read online, that motherboard is not capable of running FSB at 133 MHz.

I see, well thanks for the help anyways. I guess I'm just gonna conclude that my Kinetiz 7T also can't handle 133FSB.
But I do think some sites should be corrected, because some do claim that the motherboard supports AMD Athlon XP processors (while others claim that they don't).

I have 3 Abit KT7A mobos and
1. goes up to 138FSB stable
2. goes up to 144FSB stable
3. goes up to 150FSB stable

Plain silicon lottery: chipset cooling, motherboard revision, PSU, or recap won't make any difference.

That FSB clock is everything, over 2GHz CPU clocks (mobile-Barton) and only FSB clock makes any difference in gaming speed. FarCry and Doom3 are great for stability and performance testing.

My setup is:
-Abit KT7A v1.1
-Athlon XP mobile (Barton) unlocked (150x12,5 on boot)
-Radeon X800PRO
-ESS SOLO-1
-Aureal Vortex2
-Gravis Ultrasound
-Sound Blaster Audigy2

When in DOS I can take FSB down to 124FSB with VIAFSB and then with SETMUL multiplier down to 3x. When booting in to Windows XP multiplier is set to 14x with CrystalCPUID.

Very nice all-arounder indeed.

Sadly SETMUL freezes when I use it to change frequency. Can you share details on your board version (1.0/1.1/1.3) and BIOS? What version of SETMUL do you use?

I had some weird stuff back in the day with early multiplier set utils, so don't know if it was the platforms or the utils, like I couldn't jump down from 12.5 to 5x, I had to go 12.5 to 10.5 wait a second... 10.5 to 7 wait a second, 7 to 5. Was similar with some FSB utils also, up or down, no large bounds.

cde wrote on 2022-10-11, 22:04:

Sadly SETMUL freezes when I use it to change frequency. Can you share details on your board version (1.0/1.1/1.3) and BIOS? What version of SETMUL do you use?

See this post.

Windows98_guy wrote on 2022-09-30, 10:31:

cde wrote on 2022-09-29, 11:49:

Unfortunately I can't remember exactly my tests, but from what I read online, that motherboard is not capable of running FSB at 133 MHz.

I see, well thanks for the help anyways. I guess I'm just gonna conclude that my Kinetiz 7T also can't handle 133FSB.
But I do think some sites should be corrected, because some do claim that the motherboard supports AMD Athlon XP processors (while others claim that they don't).

Well, it depends which Athlon XP you are trying to use on KT133. Palomino's will work straight away, although with reduced clock speed as the FSB is 200 MHz on the KT133 instead of 266 that the CPU requires. I think I also tried to run 266 MHz FSB Mobile Barton on it, with pinmodding, and it did not boot at all (maybe with 1.8 GHz setting). I think I managed to pinmod a Thoroughbred to work on it with reduced clock speed.
So, I think the KT133 may have a problems at least with CPU frequencies going much above 1.4 GHz, where as KT133A motherboards generally are comfortable even at way above 2 GHz CPU frequencies. May be a motherboard model dependent issue, or there is some hard limit with some signaling speed capabilities of the KT133 chipset.

As a somewhat related thing; My QDI Kinetiz 7E (KT133A) seems to be happy with any Athlon XP (mobile Bartons included), but pinmodding is mandatory (for above 12.5x multipliers), and care has to be taken with CPU voltage settings. I am just using the latest official KinetiZ 7T/7B/7E BIOS, version "K7B BIOS 4.1", which I don't think is modified. I have some vague memory that it added some unofficial support for some Athlon XP CPUs up to 2200+. Seems to be available on this thread earlier, and I think it can be found also by searching a file named "K7B_V41LCP.zip", or using the wayback machine for www.qdigrp.com web page. However, apparently a modified BIOS would be required to have SETMUL working in MS-DOS, or using WPCREdit or pciset tool to change some register, as discussed somewhere earlier about the KT133A. Did not see a modified version of this bios on this thread after a quick search... Currently I am running a Mobile Barton 2600+ @ 2266 MHz (~17 x 133 MHz) on the Kinetiz 7E, which has been fully recapped with solid state / polymer capacitors.

Well, it depends which Athlon XP you are trying to use on KT133. Palomino's will work straight away, although with reduced clock speed as the FSB is 200 MHz on the KT133 instead of 266 that the CPU requires.

I actually didn't try my Palomino with reduced FSB. But that would defeat the purpose of it, since the idea was to have a around at least 1.4 Ghz AthlonXP, instead of regular 1.4 Ghz Athlon so it would consume less power and be more cool (would not overheat).

So, I think the KT133 may have a problems at least with CPU frequencies going much above 1.4 GHz, where as KT133A motherboards generally are comfortable even at way above 2 GHz CPU frequencies. May be a motherboard model dependent issue, or there is some hard limit with some signaling speed capabilities of the KT133 chipset.

Someone above in the post pointed out since he has 3 Abit KT133A mbs, that every single motherboard has a different FSB frequency (even though they're all the same models). I think it's because most KT133 mb don't actually reach a higher FSB, for example it only reaches 112 Mhz, instead of 133 Mhz (or 266 Mhz as they say). So yeah, motherboard model, but even than it's a lottery.

As a somewhat related thing; My QDI Kinetiz 7E (KT133A) seems to be happy with any Athlon XP (mobile Bartons included), but pinmodding is mandatory (for above 12.5x multipliers), and care has to be taken with CPU voltage settings. I am just using the latest official KinetiZ 7T/7B/7E BIOS, version "K7B BIOS 4.1", which I don't think is modified. I have some vague memory that it added some unofficial support for some Athlon XP CPUs up to 2200+. Seems to be available on this thread earlier, and I think it can be found also by searching a file named "K7B_V41LCP.zip", or using the wayback machine for www.qdigrp.com web page. However, apparently a modified BIOS would be required to have SETMUL working in MS-DOS, or using WPCREdit or pciset tool to change some register, as discussed somewhere earlier about the KT133A. Did not see a modified version of this bios on this thread after a quick search... Currently I am running a Mobile Barton 2600+ @ 2266 MHz (~17 x 133 MHz) on the Kinetiz 7E, which has been fully recapped with solid state / polymer capacitors.

That is the motherboard i actually wanted to buy, but apparantly it's quite rare, much like Abit KT7A. Also, good that you replaced the capacitors or got the motherboard with replaced capacitors. I think the previous owner of my Kinetiz 7T also replaced some capacitors with Panasonic kind.

Windows98_guy wrote on 2022-10-26, 20:27:

Well, it depends which Athlon XP you are trying to use on KT133. Palomino's will work straight away, although with reduced clock speed as the FSB is 200 MHz on the KT133 instead of 266 that the CPU requires.

I actually didn't try my Palomino with reduced FSB. But that would defeat the purpose of it, since the idea was to have a around at least 1.4 Ghz AthlonXP, instead of regular 1.4 Ghz Athlon so it would consume less power and be more cool (would not overheat).

A bit more off-topic discussion:
I just dug out my DFI AK74-EC Rev. B2 (KT133) motherboard, and did a couple of quick tests with it just for fun. Athlon XP 2500+ Thorton (basically same as Barton with half the cache), did not boot at 1.8 or 1.4 GHz (100 MHz base clock & pinmod).
Athlon XP 2100+ Palomino (multipliers unlocked and using pinmod); no boot at 1.8 GHz, but seems to be working with 1.7 GHz and under (17 x 100 MHz). Stock speed for this CPU would be 13 x 133 = 1733 MHz. Also, I modified L5 bridges on the Palomino to try to enable it as Mobile Athlon XP, but did not test yet whether I could change the multiplier with SETMUL. At least the BIOS only identifies the CPU as Athlon 4, no matter what, but that is somewhat irrelevant. The AXP Palomino consumes quite a lot of power, of course, so does not fit the second criteria...

makechu wrote on 2022-10-28, 17:17:

A bit more off-topic discussion:
I just dug out my DFI AK74-EC Rev. B2 (KT133) motherboard, and did a couple of quick tests with it just for fun. Athlon XP 2500+ Thorton (basically same as Barton with half the cache), did not boot at 1.8 or 1.4 GHz (100 MHz base clock & pinmod).
Athlon XP 2100+ Palomino (multipliers unlocked and using pinmod); no boot at 1.8 GHz, but seems to be working with 1.7 GHz and under (17 x 100 MHz). Stock speed for this CPU would be 13 x 133 = 1733 MHz. Also, I modified L5 bridges on the Palomino to try to enable it as Mobile Athlon XP, but did not test yet whether I could change the multiplier with SETMUL. At least the BIOS only identifies the CPU as Athlon 4, no matter what, but that is somewhat irrelevant. The AXP Palomino consumes quite a lot of power, of course, so does not fit the second criteria...

Interesting. This means that at least Palominos can definitely work on a normal KT133 mb, but with reduced speed.

On a side note, there are some special low voltage Athlon XP Palominos, called Athlon SFF. They have lower voltage from regular Palomino and consume even less power, but even those are rare. I get no results on ebay.

Update, 3.11.2022:
Since i had some extra free time, I decided to try Amd Athlon XP 1700+ Palomino on another board, that needs repair. With 133MHz FSB i get no post, as with the QDI motherboard, but it does work on 100, 105 and 110 MHz. It posts even on 115, but than it becomes unstable (even mb bios gives me an error). So with that i get it to run from 1.1Ghz to 1.2 GHz. All that is left now is to try and unclock the multiplier and try it to run at 100MHz FSB (14 x 100). I have also purchased Amd Athlon XP 2100+ Thoroughbred and will try it out as well in the future.

So, I went and bough a non-raid KT7A with not much thinking.
Just to confirm, the revision number is in the sticker next to the BIOS ROM? At least on my board, there was a random bunch of numbers there, with V1.3 at the end of it, so I assume this is version 1.3 board. At the bottom of the ISA-connector, there was just a sticker with text "KT7A-TIK005412", a round sticker with a "D" on it, and sticker with stamps on it, presumably related to quality assurance.

Next thing that caught my eye here is that it is mentioned that the version 1.3 KT7A uses a different firmware from earlier boards, but if I look at http://abit.ws/page/en/motherboard/motherboar … &pPRODINFO=BIOS , there is no mention of different BIOSes for different board versions, and the kt7a9.exe seems to be the newest BIOS update file. So, is this actually compatible with the 1.3 board?
If not, does somebody have the latest original unmodified bios image somewhere? At least I could not find it here on the forums after some amount of searching...
I'll try to patch the KT7A (non-RAID) BIOS file myself with the CPU microcode patches, and for enabling software CPU multiplier switching, if I just first know for sure which is actually the latest official BIOS file for the non-RAID V1.3 board.

I was a bit surprised to see some posts here that there may be some slowdowns / other issues caused by using more than 512 MB of RAM on the KT7A. I have had zero issues with using 1 GB on QDI Kinetiz 7E, which is also KT133A based, and I have all the timings set the tightest the BIOS menu allows. However, I have not done benchmark testing against using just 512 MB. Based on the 3DMark01 & 03 scores in Windows 98 SE, which I compared against what I found with google, performance was pretty close what to expect from equivalent RAM, GPU & CPU combo.
Additionally, I am using 2 TB IDE hard drives (note, using drives that emulates 4096 byte physical sectors as 512 for maximum compatibility) with Windows 98 SE on Kinetiz 7E, whereas it is said here that the ABIT-KT7A only supports LBA 28, which may limit the max. size to 137 GB. I've previously also used Quadro4 750 XGL (closely equivalent to Geforce 4 Ti 4400, but with twice the ROPs and slightly slower memory clock) with zero issues on Kinetiz 7E, but here I see that some Geforce 4 Ti models may have some issues with KT7A, which also does not sound very good to me... Currently, I have the Kinetiz 7E paired with a 256 MB Geforce 6800.
Considering these things, now I think that if I patched the BIOS on Kinetiz 7E with the CPU microcode updates and for enabling software CPU multiplier switching, I actually think the Kinetiz 7E clearly beats the KT7A for my use cases, as I have zero interest in overclocking any old hardware (although I am using a slightly overclocked mobile Barton 2600+ @ 2266 MHz on the Kinetiz 7E 😀. But, I will still do a full recap on the KT7A and see what it can do with a mobile Barton 2500+, that I have waiting on the shelf.
At least for me, stability is number one when it comes to using these things for reminiscing the good old times. And also, I mean, why keep beating up good old horses with some crazy overclocking during their retirement years while they are not dead yet :p

Finally, THROTTLE.EXE with ACPI works pretty great in DOS with the Kinetiz 7E, although I need to remember to always disable it if before turning off the computer or it hangs with a black screen. As a side note; compared to the two K6-3+ setups I also have, I find the KT133A setup to be much better with way less issues, even for DOS gaming, than either of the K6-3 platforms with VIA MVP3 (Epox MVP3C2) or Ali AladdinV (Asus P5A-B rev. 1.04) chipsets. Of course I have not tested every DOS game imaginable, but so far I have encountered many issues with the K6-3+ setups, and pretty much none with the KT133A based setup, excluding stuff that is something purely sound card or video card related.